Research On Supply Chain Network Optimization Problem Based On Single-Objective And Multi-Objective Intelligent Optimization Algorithm

In the context of economic globalization,supply chain management has become a critical strategic tool for enterprises to enhance competitiveness.In the field of supply chain management,many complex supply chain network optimization problems need to be addressed.These problems are typically characterized by a vast network structure,multiple conflicting optimization objectives,and complex constraints,which increases the difficulty of problem sloving.Intelligent optimization algorithms have become one of the essential tools for solving supply chain network optimization problems due to their adaptability,robustness,and excellent global search capability.This paper focuses on the modeling of complex supply chain network optimization problems and the design of corresponding intelligent optimization algorithms,with the aim of providing more effective solutions for supply chain management.Additionally,the paper explores the field of many-objective optimization,aiming to provide innovative ideas and methodologies for addressing many-objective supply chain network optimization problems in the future.The main research of this paper includes:1)The Green Supply Chain Network(GSCN)model with complex constraints is designed in this paper,which focuses on pollution emissions during the supply and manufacturing stages.The model introduces a novel green constraint aimed at reducing pollution at the source.The objective of this model is to minimize the total operational cost of the enterprise and satisfy the given constraints by optimizing logistics routes and product freight.To efficiently solve the GSCN model,this paper proposes a Marine Predators Algorithm with Stage-based Repairment(MPA-SR).The stage-based repairment operation consists of a path decision operator and a product freight allocation operator.The path decision operator adopts a random strategy to plan legal logistics routes,enhancing the diversity of the population.The product freight allocation operator allocates product freight to logistics routes based on a greedy strategy to optimize the solution quality.Additionally,to balance the performance and efficiency of the algorithm,an adaptive parameter for controlling the repairment probability is proposed.2)The Constrained Large-scale Multi-objective Supply Chain Network(CLMSCN)model is designed to simultaneously optimize the total operational cost and customer satisfaction under capacity constraints.To effectively solve this model,this paper proposes a Coevolutionary Algorithm based on the Auxiliary Population(CAAP),which utilizes two populations in a collaborative way to optimize the problem.The first population is used to solve the original complex problem,and the second population is used to solve a simplified problem with no constraints.This approach can significantly enhance the global search capability of the algorithm.In addition,to address the original complex problem,this paper designs a linear repair operator that can improve the feasibility of infeasible solutions and accelerate the convergence speed of the algorithm.3)To address many-objective optimization problems,this paper proposes a Nondominated Sorting Local Search algorithm based on the Clustering Entropy Selection(NSLSCES).To increase the selection pressure of the algorithm in high-dimensional objective space,the Clustering Entropy Selection(CES)method is introduced.This method employs a variant of information entropy,clustering entropy,to effectively evaluate the distribution of the population in high-dimensional objective space.The CES method is based on the principle of minimizing Gibbs free energy and uses an individual absolute energy calculation formula based on angular distance to describe the free energy of the population.Furthermore,to reduce the time complexity of the selection process,the paper adopts the component thermodynamic replacement strategy,which improves the accuracy of selection while reducing the time complexity.